The comparison
| Type | What it senses | Accuracy | Drift | Response | Used for |
|---|---|---|---|---|---|
| Capacitive | Polymer film capacitance | ±2–5% RH | ~1%/year | Seconds | Nearly everything |
| Resistive | Salt or polymer resistance | ±5% or worse | Faster | Tens of seconds | Low-cost devices, HVAC switching |
| Hair tension | Hair length | ±5–10% | Substantial | Minutes | Antique and decorative dials |
| Psychrometer | Evaporative cooling | ±2–3% | None inherent | 1–2 minutes | Field reference, calibration |
| Chilled mirror | Dew forming on a cooled surface | ±0.2°C dew point | Minimal if kept clean | Seconds to minutes | Laboratory reference standard |
| Gravimetric | Mass of water extracted | Primary standard | Not applicable | Hours | National metrology institutes |
Reading the table
Accuracy is not the whole story. A capacitive sensor at ±3% that has been on a shelf for four years may be further out than its specification suggests, while a psychrometer at ±3% will still be ±3% in a decade because nothing about it drifts — it is a thermodynamic relationship rather than a calibrated material.
That distinction is the main reason psychrometers survived so long in professional use, and why they still appear in calibration procedures.
Response time matters more than people expect. A hygrometer that takes ten minutes to settle is useless for walking around a building looking for a damp spot, however accurate it eventually is.
Drift is what actually ruins consumer readings. Roughly a percent a year for a decent capacitive sensor, faster for resistive, much faster for hair. After five years, an uncalibrated device may be well outside anything it claimed on the box — which is why the salt test is worth twenty minutes of your time.
What each is actually for
- Capacitive — everything, unless you have a specific reason otherwise. Cheap, fast, stable enough, no maintenance beyond calibration.
- Resistive — switching something on and off, where a few percent does not matter.
- Hair tension — an object on a wall. Charming, historically important, and not a measurement instrument in any modern sense.
- Psychrometer — checking whether your other instruments are lying, and any situation where a traceable reading matters more than convenience.
- Chilled mirror — laboratory and industrial work where the dew point itself must be known precisely. Expensive, and the mirror needs cleaning.
- Gravimetric — how the entire chain of calibration is ultimately anchored. Not something you will operate.
The one that is not an instrument
There is a seventh source of humidity readings and it belongs in any honest list: a weather service.
It is not a mechanism but a network — real instruments from the rows above, distributed across a region, combined with model output and interpolated to a point. That is what a forecast, this website's live pages and the app all report.
Its error profile is completely different from a handheld instrument's. Sensor accuracy is excellent, because official stations are well maintained; but the nearest one may be miles away, and distance is the dominant error. It is the right tool for outdoor questions and no tool at all for indoor ones.